Positional-aware Spatio-Temporal Network for Large-Scale Traffic Prediction

Researchers propose PASTN, a lightweight neural network for large-scale traffic flow prediction that uses positional-aware embeddings and temporal attention mechanisms. The model demonstrates improved efficiency and effectiveness across various geographical scales from counties to entire states.

• →PASTN introduces positional-aware embeddings to better distinguish individual nodes in traffic networks.
• →The model incorporates temporal attention modules to improve long-range historical pattern recognition.
• →The lightweight design addresses deployment challenges in real-world applications with large datasets.
• →Testing across multiple scales (county, megalopolis, state) validates the model's scalability.
• →The end-to-end approach effectively captures both spatial and temporal complexities in traffic prediction.